Native AI Product Manager
Yuri Gomes de Abreu
I build AI-native products — and the agentic systems that run them.
10+ years across Mercado Libre, Uber, Apple, and McKinsey, building marketplace, pricing, and platform products. Now designing agentic product systems where AI handles the execution and people make the calls.
Open to Director / Group PM / Principal PM / Head of Product roles (US · UK · EU; open to relocation) — and a small number of fractional & advisory engagements.
Background
- 10+
- years across Mercado Libre, Uber, Apple & McKinsey
- 3.5M+
- marketplace participants served
- 15
- countries of product scope
- ~75%
- of routine PM execution automated with agentic AI
- 4
- languages (EN · PT · ES · FR)
About
Senior product leader with 10+ years building marketplace, pricing, and AI-native products at Mercado Libre, Uber, Apple, and McKinsey. I've owned pricing and platform products serving 3.5M+ marketplace participants across Latin America and global markets, and I build the agentic systems that run the execution.
Today I design product-management "operating systems" where LLM-powered agents handle the execution and humans make the calls — pairing engineering depth and hands-on AI fluency (Claude, GPT, Gemini; RAG, MCP, evals) with a Wharton MBA.
I'm focused on Director / Group PM / Principal PM / Head of Product roles at AI-native companies. Based in Brazil; open to relocation (US / UK / EU). Fluent in English, Portuguese, Spanish, and French.
How I work — AI-native
I work AI-native. Instead of doing routine product execution by hand, I build agentic systems — LLM-powered "skills" for issue scoping, dev kickoff, QA, PR review, lifecycle communications, and support triage — so the repetitive work runs itself with human approval gates, and judgment goes where it actually matters.
On a recent engagement this approach cut a product team's steady-state execution load by roughly 75% (from ~20 to ~5 hours/week) while raising consistency and speed. I'm hands-on with RAG, MCP architectures, evaluation frameworks, and multi-model workflows (Claude, GPT, Gemini), and I prototype and ship in code.
Built with AI
Products I designed and built end-to-end, AI-native, by myself.
Lab
Side quests that keep the range sharp.
Interactive 3D knee anatomy
A real-time 3D anatomical model rendered in the browser with Three.js.
Three.js · WebGL
3D modeling with AI
3D modeling and scene work driven through Blender + MCP tooling.
Blender · MCP
DIY smart ventilation
A home ventilation system built on ESP32 with ESPHome and Home Assistant.
ESP32 · ESPHome · Home Assistant
Where I've worked
Four countries on the ground — product scope across 15.
Experience
- Own pricing and shipping product strategy for the LatAm seller ecosystem.
- Led a large-scale seller shipping-pricing redesign across millions of transactions.
- Shipped 15+ pricing changes with measurable improvements in quality and revenue.
- Built agentic AI workflows that auto-prioritize the product roadmap.
- Drove 3x revenue growth and expanded gross margin to 30%.
- Structured Finance, HR, Product, and Customer Service functions; national Top 3 customer-service award.
- Grew EBITDA 15% and total-factor productivity 10%; reformulated business strategy and org structure.
- Raised customer-service level 40% and quality 20%; designed an internal ERP and HR platform for 100 employees across 5 countries.
- Led cross-functional delivery of the new Apple Music Messaging vision.
- Evaluated and launched ML/AI personalization features improving user engagement.
- Extended Apple's official project-lifecycle framework with a new program phase.
- Owned pricing for 3.5M drivers across 15 LatAm countries.
- Built Uber LatAm's first Python pricing-simulation platform (5-day decisions → 1).
- Captured $5M in inefficiencies in four months; raised driver satisfaction 16%.
- Owned a proprietary consumption-forecasting platform and team across 54 projects, 35 clients, and 6 sectors.
- Fast-track promotion one year ahead of standard.
- Implemented a knowledge-management system, increasing innovation pace 20%; contributed to a flagship product launch.
Let's build something.
Book a 30-minute call, or reach out directly.